Recurring Failure Escalation Service

Problem Description

A media streaming company runs a transcoding pipeline that processes user-uploaded videos. One particular transcoding worker occasionally fails with a "codec mismatch" error on certain file types. This error is not immediately catastrophic — the job can be retried — but the same failure sometimes persists for hours or even an entire day without being noticed because each individual retry looks like a new event to the monitoring system.

The on-call engineering team has been burned by this pattern twice: a "codec mismatch" failure quietly retried for 18 hours, consuming compute budget and delaying user content, until a human noticed. They want a failure tracking module that remembers the history of a recurring failure type and automatically escalates when the problem has persisted too long or recurred too many times, rather than treating each attempt as fresh.

Output Specification

Write a Python module recurrence_tracker.py that:

• Tracks failure recurrences by a failure key (string identifier)
• Decides whether to suppress or escalate based on recurrence count and elapsed time
• Clears the tracking record after escalation
• Includes configurable thresholds (MAX_RECURRENCE and MAX_WINDOW)

Also write a demo_tracker.py script that simulates a recurring failure being checked multiple times (at least 5 checks) and prints the suppress/escalate decision at each step. Use mocked timestamps to demonstrate the escalation trigger (do not rely on actual sleep delays).